> AI agents: this is one page from PostHog's docs. Full index of Markdown docs for LLMs: https://posthog.com/llms.txt # v0 integration Canonical URL: https://posthog.com/docs/integrations/v0 > **Cloud region:** This page uses US Cloud hosts. For an EU Cloud project, use `https://eu.posthog.com` instead of `https://us.posthog.com` (app host for links). ## On this page - [Connect PostHog to v0](#connect-posthog-to-v0) - [Set up PostHog in your project](#set-up-posthog-in-your-project) - [What you can do with the MCP](#what-you-can-do-with-the-mcp) - [Troubleshooting](#troubleshooting) - [Next steps](#next-steps) The PostHog MCP lets v0 query your analytics data, investigate errors, manage feature flags, and run experiments - all through natural language. ## Connect PostHog to v0 1. In a v0 chat, click the **+** button in the bottom left 2. Click **MCPs** 3. Click **Add MCP** 4. Select **Custom MCP** 5. Enter the name `PostHog` and URL `https://mcp.posthog.com/mcp` 6. Under **Authentication**, select **OAuth** 7. Click **Add**, then authorize with PostHog when prompted The PostHog MCP server is now available in your v0 chats. ## Set up PostHog in your project Once connected, give v0 this prompt to add PostHog with the Vercel Flags SDK: ```text ## Add PostHog Analytics and Feature Flags ### Packages `posthog-js`, `flags`, `@flags-sdk/posthog` ### Critical: Let the Client Own the Distinct ID **Do NOT generate distinct IDs server-side.** The PostHog JS SDK manages its own `distinct_id` via `localStorage+cookie` persistence. If you bootstrap a server-generated ID, every server render can produce a different ID, breaking experiment attribution. ### Implementation Structure - `/components/posthog-provider.tsx` - Client provider that initializes `posthog-js` with `persistence: "localStorage+cookie"`. Guard against re-init with `if (posthog.__loaded) return`. No `bootstrap` option needed. - `/app/layout.tsx` - Wrap children in ``. No async work or distinct ID generation needed. - `/lib/flags.ts` *(optional, for server-side flag evaluation)* - If you need server-side flags, read the distinct ID from PostHog's native cookie `ph__posthog` (JSON format: `{"distinct_id": "..."}`). **Never generate a fallback UUID** -- if the cookie doesn't exist, skip server-side evaluation and let the client handle it. ### Reading Feature Flags Client-Side Use `posthog.onFeatureFlags()` to wait for flags to load, then `posthog.getFeatureFlag("flag-key")` to read the variant. Always wait for flags to be ready before rendering flag-dependent UI. ### PostHog MCP Notes - The MCP may default to a different project than the one your API key sends events to. Use `PostHog_projects-get` to list projects and `PostHog_switch-project` to point at the correct one before querying. - Use `PostHog_query-run` with HogQL to verify events are flowing. - Experiments created via `PostHog_experiment-create` use `control` as the default variant key (not the variant name). Map `control` to your first variant in code. ``` The agent will set up PostHog with the client owning identity via `localStorage+cookie` persistence, ensuring reliable experiment attribution. ## What you can do with the MCP Once connected, you can ask v0 to interact with your PostHog data directly: ### Query your analytics **Prompt:** "How many unique users signed up in the last 7 days, broken down by day?" The agent will run a trends query and return daily signup counts with a breakdown. **Prompt:** "What's the conversion rate from signup to first purchase?" The agent will build a funnel query and calculate the conversion percentage between those two events. ### Investigate errors **Prompt:** "What are the top 5 errors in my project this week?" The agent will fetch error groups sorted by occurrence count, including how many users were affected. **Prompt:** "Show me details about the most recent TypeError" The agent will pull the stack trace, affected users, and browser/OS context for that error. ### Manage feature flags **Prompt:** "Create a feature flag called 'new-checkout' enabled for 10% of users" The agent will create the flag with a 10% rollout and return the configuration with a link to the flag in PostHog. **Prompt:** "Update the 'beta-features' flag to 50% rollout" The agent will modify the existing flag and confirm the new rollout percentage. ### Run experiments **Prompt:** "Create an A/B test for the pricing page that measures conversion to checkout" The agent will set up an experiment with control/test variants and configure a funnel metric measuring pricing page to checkout conversion. See [all available MCP tools](https://posthog.com/docs/model-context-protocol/tools.md#available-mcp-tools) for the complete list. ## Troubleshooting ### OAuth not refreshing If the OAuth token expires and won't refresh, you can use a personal API key instead: 1. Create a [personal API key](https://us.posthog.com/settings/user-api-keys) with the **MCP Server** preset 2. In v0, go to **Manage MCPs** and edit the PostHog connection 3. Change authentication to **Bearer** and enter your API key ### MCP not connecting Try disconnecting and reconnecting the PostHog MCP server. Make sure you're logged into the correct PostHog account. ## Next steps - [Vercel Flags SDK docs](https://posthog.com/docs/libraries/vercel.md) - Full Vercel Flags SDK setup guide - [MCP tools reference](https://posthog.com/docs/model-context-protocol/tools.md#available-mcp-tools) - All available MCP tools - [Feature flags](https://posthog.com/docs/feature-flags.md) - Set up feature flags in your code Have feedback or questions? Let us know on [GitHub](https://github.com/PostHog/posthog/issues) or [ask a question](https://posthog.com/questions.md).